Alan Kay SCIx Keynote Presentation (2012)

From Viewpoints Intelligent Archive
Revision as of 19:28, 25 November 2021 by Ohshima (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
good evening
I'm Chris Johnson I derived the scientific
computing and Imaging Institute here at the University of Utah better
known as the ski Institute and I
want to welcome you to our second ski X open
house and to the to our keynote presentation
this room welcome especially
for people who are from outside of Salt Lake City
of Utah who have come here from all over the world
as part of this info conference
and welcome also the people from
Utah so it's my great pleasure to introduce Alan
Kay and one of our University of Utah alumnus
and he is one
of the true innovators in computing we
were just talking just now about there were
fables about his dissertation when he was here as a graduate
student and I was getting
some clarification on what really happened so he was one
of David Evans PhD students and
David Evans view was that you
you needed to do first-class research in
order to to prove yourself and get your union
card and move on to do interesting research and
Alan was doing a lot of
interesting research and if you look at his dissertation it
includes hardware design a software design
network design user interface design
and as some of the elements from which he
was able to move on and do some fundamental research
in object-oriented programming and hardware
design and interface design with the Dynabook
with with computer
engineer manufacturing with
many many things that that he
worked on since he left the University
of Utah so along the way he was at Xerox
PARC for about ten years after
that he became a Tori's chief scientist for
three years some of you may know that Nolan
Bushnell was here just last week and
so we were able to hear about some of the
times that's at Atari from Nolan about
ten years ago he founded
a nonprofit organization called viewpoints Research
Institute in which he's continuing
his work on cognition and learning and
programming or with children
Allen has received a number
of awards and maybe I should just note the
2003 Turing Award for
his work in object-oriented programming and the
creation of the Dynamo so please
join me in welcoming Alan Kay thank
so the aim of this talk is to
give you a sense of
what the research community
was like from
whence all of the stuff we today we have
came from this was
suggested by Chris based on a a an
essay I did in
as a tribute to the community when several of
us won the Draper prize and we
all felt that somehow the community
should be getting these awards and I'll try and give
you a sense of that because Utah salt University
of Utah was part of that committee in
one of the Thornton Wilder
plays he has an old
fortune teller one
mechanical fortune tellers and she says I tell
the future it's easy but telling the past is
really hard and the reason is is the
past is so detailed and
a lot of the details actually are important
another part of it as all
of us discovered when we
were asked to write histories some 20
25 years ago is that
many things that
many pathways are taken
are taken not just
because some idea beckons but
also because one has been disgusted
by work
one season round and for instance IBM
back in those days was a constant source
of inspiration to do better and
this makes
telling an honest history
almost impossible because you can
praise somebody in a sentence so-and-so
was great and we got some ideas from them but
when you do the opposite
of praising somebody was saying this work was so bad
that we decided to do this you can't just
do it in a sentence it's not fair you have to
explain why and so the histories we all
wound up writing emitted many
of the main motivations for some
of the the things that we did but so
first I'll give you a little outline of
how this worked some of the people here
I think are old enough to have been alive
when when this stuff started and
this slogan goes
back to the world
two radar effort at MIT
building 20 radiate
radiation lab and many
people don't realize that radar was actually
the technology that won World War two force
because we're in Dane danger of going
under from German u-boats the
British had invented more
radar than we had people were fooling around with
it and for
a variety of reasons they did not have the resources to develop
it so they came over not
just with the ideas but with
the fundamental technology that made radar packed
practical which is a device called a cavity
magnetron yeah and gave
it to MIT
and if you
think about world war ii remember December 7th 1941
was in December so
41 had already happened so it was 42
43 44 and a half of 45 so world
war ii was a big deal but it was only
three and a half years compared to what we're used to
today of things that go on for a long time
and this group of
people in building 20 turned out
more than 150 different radar
systems invented
engineer packaged
and had them built by the route 128
companies around Boston and installed on every
kind of ground installation
seagoing and air going vehicle so
it's one of the most amazing bursts of
combining science and engineering
done in a short time to
make things that actually actually working
were effective and another
somewhat similar effort was at
Los Alamos working on the
atomic bomb all of the atomic bomb
was not nearly as critical to ending
the war as as radar was so
part of the
both of these groups were funded
by President Roosevelt's science
advisor Vannevar
Bush and Bush had been an MIT
professor and had done some of the early analog computers
and so you can think of this as kind of an old boys club it
really was and these are pretty
good old boys
they're basically gentlemen
and then right after World War two
Cold War and one of the big efforts there was
to try and deal with threats of
Russian bombers
and so this incredible system called semi-automated
ground environment was made those
are displays that those people are peering
into the air and enormous computers
were built
and they were built by many of the same
people some of whom
worked on the radar were now sort of in more of an administrative
fast for instance Jerome
Wiesner was an undergraduate at
MIT during World War two and
later was a head of the department
of engineering then later became president MIT
and then later President Kennedy's science
advisor so these
these things go on and on meanwhile
Bush at
during the 50s created
NSF in order to deal
with a little bit more
look ahead than we had in World War two and
in the 60s
actually in the late 50s ARPA was created
because of Sputnik to do something about
what was perceived as a great threat and I'll
tell you a little bit about one division of it that
created in the early 60s called the information processing
techniques office from which
it's not an exaggeration to say most of the technology
we have today got its initial funding
MIT was the first recipient
Carnegie Mellon and in
1965 the University of Utah so they're
about 15 places two-thirds of them universities
and about a third militarily
defense related companies like mitre RAND
Corporation systems Development
Corporation in and so forth and
these places created
an enormous number of inventions and
part of the inventions they created were
PhDs over the six
years or so this was operating and those PhDs went
in the 70s to Xerox PARC and did
personal computing as we know it today so
this is kind of a oversimplified capsule
summary but it's basically this
idea of making progress is more important than anything
something that was carried through in each one of these generations
it was the secret to
making this work because of course being
humans and anybody who's ever spent
any time around a university and gone to faculty
too well how easy it is
for people who have
gotten awarded for being rational to
be irrational in a meeting and
so and of course there were rivalries
and everything else and the thing they decided in
the rad lab was they didn't have time for any
of that so somebody
nobody knows who came up with this ideas hey well we're
is the only thing we're interested in forget about everything
else that is all we're going to work
on that gap passed into the 50s that got passed into the 60s
that's what I learned when I came here into the ARPA
projects here and that was what was
carried into park and the thing that none
of the histories of
these efforts
has really gotten clear enough
I believe is why why
did they work it was because the amount
of cooperation that actually happened
would be remarkable I think to almost anybody
today it's just hard
to believe just
how open how
everything was shared and it was just part of the part
of the deal and so
when you have cooperation amongst really smart people
then you get synergy that
is out of this world you can
get 185
different radar systems in a couple of years you
can get nuclear
fission in a couple of years you
can get enormous
computers the size of football
fields running on vacuum tubes that never crashed
in a couple of years it's
just unimaginable and of course I can't tell you
about all of it but I'm just trying to get you into
this era where
the we basically
live in an era today where the components are
really reliable and the people are less so and
back then it was exactly
the opposite the components are like the worst things
you ever if anybody here old enough to do vacuum tubes
and the 50s 60s yeah
so it
was just incredible
doing that but because everybody
cooperated in a way that is very difficult to explain
wonderful things happen so here's
the first exhibit here and
I'll direct your attention to where
it says second floor there
and it says computer a and computer B
and computer
a is the size of a football field and
so is computer B
so that second floor is the size of two
football fields so it's about two acres of
vacuum tube computer
down below is all of the power and
stuff for it up top
so when I when I visited one of these in
the I guess
in the late 60s at Systems Development Corp
they took me into a room that was bigger
than this but it's like a typical
glass house mainframe machine room and
lots of noise and raised floor and all that
hey said well what are you thinking I said well I
thought it was somehow I thought it was larger than this they
said well you don't understand this is just the console
it went over this is just like out of a James Bond movie they
went over and pulled these drapes and down
and we were up on the third
floor there and you could look down onto
these double football fields of the
twin Q 32 s there
and there are wonderful stories
about this machine I'll just tell a couple
first first interesting
one is the last one of these
did not go out of commission until the
early 80s and
so some of you might be wondering well
where did they get all these vacuum tubes from
knows to tell us yes
isn't this great yes
as these systems went on and on
we we ceased making jet vacuum tubes
Japanese he's making vacuum the Russians are still making vacuum
tubes so we started buying vacuum tubes from
them in order to put into these this
defense system against them
can't beat it the other interesting thing
was that because they didn't want
the thing to crash they
did several things one is they had two computers running the same computations
and checking them so they had a kind of a corpus
callosum between them but
they also the part of
this huge room that was just the console where
other computers whose job it was to do Diagnostics
on the big computers while they were running and whenever
they detected an instruction failing they would patch
in a subroutine that emulated that
instruction that was made up from the instructions that were still
working so one of these babies
started crashing it took like five or six
days we just get slower and slower
and pretty soon it was almost all software patched
in there's maybe two or three instructions and they
always were able to fix these machines before they completely
went went down in spite of the fact that
they would be having back in tubes going
out every few minutes on these
and you could see that display screens
that they had
on them and that thing the guy is holding there as a light gun
so the way
graphics was done in those days the
segments there actually plotted points
were plotted in sequence and
could register when it saw one in the computer knew
when it drew that point and what it belonged to and would cause an interrupt
and so forth okay so that's from
the 50s so here's another thing from
the 50s this is John McCarthy as he looked
back then and John took one look at
these sage consoles and said well everybody's gonna have one of those in their
house before long and so
his idea was because we've already got
elected electricity
utilities we've got water utilities
we've got gas utilities why not an information utility
what they call it back then so this is a cloud idea
the first time around and
he started thinking about that so he wrote this memo
in the late 50s to the
provost at MIT
trying to get him interested
in paying for
doing a time-sharing operating system for
new computer they were going to get from from IBM and
one of the things he says in it suppose that the programmer has
keyboard the computer then he can try his program interrogate
individual individual pieces of data or program to
find an error and make a change in the source language and try again
ability to check out a program immediately after
writing at save still more time and so on now
actually if this were a history lesson about
back then instead
of about Utah I would show two other
pieces of writing that John did
in the 50s also one of them called
for an artificially intelligent agent
to serve you through
the terminal and the other one called
for the invention of a programming language now
called Lisp in order to program
artificial intelligent agent and John by the way was the guy
who made up the term artificial intelligence but
it came actually from thinking about what
future this guy was
smart and even a few years before
that here's Dave Evans and Harry husky
dave was a an executive
at Bendix and at
various times on this project he was a
staff engineer and I
wound up actually running this project but
this quite a few of these but I
would call this a personal computer wouldn't you
quite a few of these G 15s
were hairy husky by
the way he had worked with touring on the pilot ace in England
and later went to Berkeley with with
Dave Evans so not only had
these acres size machines we had people thinking about
what does it mean to have your own machine
okay I'm gonna pick the Year
62 here now we're in the 60s for a variety
of reasons first it's 50 years ago this year
that's a nice round number second it's
when I started programming my first programming
for money was on this machine the IBM
1401 maybe some 1401
programmers here it was a bit of an odd machine to
put it put it mildly but in many ways it
was completely utilitarian in
the purposes that IBM had
put it out which was to replace
without doing any psychological
damage to anybody these enormous
punch-card accounting machine floors
that's what it was for so
wrote programs to do what the FAK you work from the actual flow
charts back then I was in the air force
doing this so
you think computing is mundane today well it was
mundane back then and that's because they're just a lot of money
people in computing
and partly
because business is so Monday they
never look into the future they always want to find out what happened
in the past so they're constantly doing
accounting instead of doing something like
making probes into the future but
oh well ok so that
is our mundane slide however
there's also the 50th birthday
of what
most of us think is the world's first real
graphics system first
real personal computing system and I've
got a little movie here I hope you can see it looks
a little dim but let's see
so the reason it's twinkling
like that is it's plotting individual points and
to keep you from
getting sick they would be randomized
in memory so about half of
this this is done on one of these acre-sized machines
notice it has a clipping
ability and
if you notice it straightened up that
flange now it's going to straighten up these lines
automatically for you now
he's asking it to be collinear so
he draws these lines they're going to lay down right
on top he's using
those full lines he did as guidelines
for these dashes now he's going to make the guidelines
transparent he's got a hole
in the flange
sketchpad doesn't know anything about
flanges or rivets
going to make a rivet to put it in the flange here and
notice he can just sketch
without trying too hard it's
going to use that as the center for that arc because he can point
at things and tell sketchpad to solve a
make all those edges mutually perpendicular and there
you see sketchpad solve that problem
it's a nonlinear problem
and can do many solutions
of course he could constrain it so
that the
edges are fractions
of each other and so forth so sketchpad
had three problem solvers could
solve fairly complicated problems now here's another interesting thing this
is an instance of that master
rivet that he just drew you can
see it can be rotated and scaled he's gonna
lock it into the flange there
little flick and
now he's going to show us he could make some
more instances of those rivets so
this is the first object-oriented system that
I know of he forgot to get
hese Crosse things so he goes to the Masters and makes
them transparent and so the rivets feel
that change
and anything that he makes he can make into
into a master
so he's made that into a master
and now he's going
to make instances of that thing he just drew
pretty cool huh so
what sketchpad had
here was real
time interaction it had
masters and objects which we
call classes and objects it had this powerful
problem-solver so
it was programmed not in terms of
procedural programs but in terms
of what you wanted the results to be and
it could do some fairly hard problems like this
bridge a problem here
now the bridge is something we don't have a movie up because this
acre-sized machine was so slow
it would take the bridge about 30 seconds to
solve and it was too slow to make a movie of so in
honor of I would say ask Ivan
by the way how could you this is Ivan's Ivan's
PhD thesis I
still think it's the greatest single
PhD thesis done in computing and
I said huh Ivan how could you have possibly done
the interactive graph he had to program the display
sakes it was just that's just an oscilloscope
that's being used there how
could you have done that the objects and
the problem solvers all by yourself
in one year and he said well I didn't know
it was hard
so in honor of the
of this 50th birthday
we actually have recreated
sketchpad here so
we have the twinkling here and
I can turn on gravity
and of course my my little Mac
Pro here is thousands and thousands
of times faster than the acre-sized machine back then
so watch what happens to notice this
little beam is saying
it has zero stress and strain on it but if I if I
drop it in here notice that not only does it change
but the entire simulation feels
it similarly I can do this here
sketchpad didn't know a darn thing about bridges
is part of its charm so and on some weights here
and of course one of the
things that this led to was the desire for nicer displays
and we certainly did
and of course if you think about the you know looking for the keys under
the lamppost instead of in the inconvenient dark place you
lost them when you're faced with something like
sketchpad you could either make sketch
pad better which means you're
operating in eivin's range or you could make the display better
so virtually everybody in computing decide
to make display is better so we have better looking
display here's here
it is a modern time we can do
a few few more tricks here so for
instance we can say thanks Ivan
words have weight
we did the we did the constraint
solving a little bit differently but basically this whole
bridge simulation just has three lines
of code in it and here
they are they're in the the
bottom weight they're the top
one here is Galilean
gravity constant acceleration gravity this
is the spring constant so one way two simple
way to think about steel is that it
is actually springy you've ever seen
the Tacoma Narrows Bridge come down
it's brought home right right away and so
just putting a vector
strain constant on every every
line in here will do it and then this bottom guy is
the pins that are holding the beams together so
if I get rid of the top guy here
gravity is turned off
and this whole thing should settle down to zero
because just being being drawn by the string
forces okay so I turn it back on and
now we have to think about what
will happen if I undo the pins
okay this is just to make the
point that what you see on the computer
screen is just a costume what's interesting is
what's down below and
we generally don't get to see what's down below the costumes
seem to have fixed purposes here and
I'm gonna repurpose all of these beams
into the user interface of the system I'm
giving this this demo in
I don't
really like this steel look here a lot of Steve
so we could try a different color here that's
to Microsoft like
and so this is this is actually the system I'm
giving this talk in here's the sketchpad thing and
it's Turing slide here I'll go back
up to full screen here the stirring slide here is to remind
me to to tell you
that one of the things that was happening around
this time was thinking about biology in
computational terms and vice-versa
some of us had degrees in in
in biology and
Turing was one of the first to work
in both areas and some of us I in fact
read Turing's paper on morphogenesis before
I read his paper on on
Turing machines and one of the
interesting things about biology is the way systems work so here's some
ants and you notice when they wander into food
there they start putting out a perfume which starts
diffusing and one ant wanders
perfume and it doesn't have food it goes immediately upstream
and finds the food very efficiently and
then goes downstream so now all the ants
are occupied as though they've
programmed in some coordinated fashion but
in fact they're all working independently
so this is highly parallel programming
here organized by
sending messages
so we call this particles and fields and
it's something that works everywhere
there are people in in the skee x
exhibit that we're studying some of these
how biological gradients work and
once you start thinking about computing it is very
very difficult to not
start thinking about the system's properties of
biology and this community that I'm
telling you about certainly did
another thing that happened in 1962
was this machine which
is since it has a display is
probably the first machine ever to have all the essential
attributes of a personal computer and
the fun part of it is that this guy Wes Clark was
the architect of the acre-sized machine
that Ivan did sketchpad on and also
did this one for biomedical engineers because
couldn't wait for the University mainframe they want
he wanted to have something in the lab and the fact the first 20
of these were built by these biomedical
technicians Wes organized this machine
into a kit and in order to get
one of these machines you had to go to Lincoln labs and prove you could build it
and that meant you could take care of it when you took
it back so by the way about 2000
of these Linc machines were actually built
and used in the in the 60s so
this is not onesies or twosies another
thing that happened in 62
was the B 5000 by Bob Barton
came out and this this
is still one of the most advanced piece of hardware
ever done more advanced than any of the chips
that we have in our personal
computers or iPads and stuff like that this guy was a
complete genius and we'll meet him again pretty soon
1962 was when angle
Bart sent his proposal
into the
DoD complex saying
we need to augment in human intellect in computers are
the way to do it and this is a great picture of him because
he was very much Moses wanting
to lead the children of Israel out of the
darkness of Egypt this was
his personality and
then things got started
because they had already been started
and I think people have studied the history of the Renaissance
know that the Renaissance happened before the printing
press it antedated the
printing press by 60 or 80 years
so that when the printing press happened
it happened in something that was already starting
and it provided an enormous acceleration
what happened here with Licklider Licklider
was a psychologist who'd gotten interested in computers
he said
the destiny of computers is
become interactive intellectual amplifiers for all people
universally networked worldwide whenever
they asked him what he was doing with this millions
of dollars he had been given to pass out
this is what he would say and
here's a memo he wrote a year
later in 63 to the members and the
affiliates of the intergalactic computer network
asked why are you calling it the intergalactic computer
network he said well engineers always do the minimum
says I can't get them to realize
that I want to connect every person on the
planet so I've been calling it the intergalactic Network
to force them out of doing it the way AT&T
might try to do it
and of course we have that network today he funded
it and here's the nice thing he said if we succeed
in making an intergalactic network then our main problem
will be learning to communicate with aliens and
what's interesting about this is we gave
him the first thing but computing
has been terrible at giving him the second because the aliens
means not just communicating with other people means communicating
with other software means software over here
communicate with other software once you scale this thing
up you have an enormous
communications problem which
most people have been afraid to
work on the last 30 years
and another thing that Lickliter
decided was that you couldn't think of
good goals inside the beltway in Washington so
he said I'm gonna fund people not projects so
lot of these people have actually won the the
Turing award the equivalent of the Nobel Prize
and computing and fact all of these this
just from the first five years this is for up to
about 1965 and quite a few
more people who are funded by this funding
agency and want it since
so this is a real picture
I will not tell you where it
was taken the sign there says do not
touch any of these wires
this is one of the most honest slides I've ever
seen about how most people do things
and of course software is much worse
because this might have only 10,000
wires and software has hundreds of millions of
lines of code which is tangled very much in this
same way and it's basically
the tendency of humans to be tactical
rather than strategic and to do
things incrementally because it seems rational
you know we won't do something
radical we'll just adapt
what we've got and that doesn't that sound reasonable yeah
except it doesn't scale and
you wind up with this awful stuff here and
a simple way to characterize this
community that Licklider
was funding these 15 universities and
companies was that
they were exactly the opposite of the way IBM
thought about doing things in almost every respect
and it was also a much smaller
group of people and they had much bigger ideas and
so in order to do that you something has to
give and basically they decided
that they would give up on performance in
order to get system integrity and
this they didn't have a meeting to decide on this
it just was the thing that was happening I
saw it happen when I was when I was
here starting in 66 and so
the way they would think about this massive
wires is Wow this is just
a communications channel with everything
that wants to be communicated with hung on
it period that
is what it is and we'll we'll make different
kinds of these and this will give us all
the crossbar combinations that we
that you could possibly do but it'll be simple
it just has the the problem if the first order
is a little bit inefficient because the
you have channel contention of one kind or another but
this is just the way this group
thought about things and this
idea that this is a recursive idea
that you can extend all the way down
of the layers of software all the way outward through
all of the layers of networking is something
it was completely out of the way IBM
Burroughs and the other vendors thought at that time
so mid sixties here's
Dave Evans at Berkeley and Butler
Lampson one of the most brilliant people any of
us have ever met it's
frightening to realize the butler wound up as being the principal
investigator for ARPA on this project in
1965 when Dave left to come here in
a 1965 Butler
was 22 years old and he
was always a senior
you know Butler from the moment he lit
his eyes on a computer he got in a high
degree in physics at Harvard and came
out to Berkeley to study physics and went through the wrong door as
he put it and found
this project to do you know a lightweight
time sharing system that could actually be practical
he never came out of that room
and so this machine was called Project
genie back there but it became
a commercial machine called the SDS 9:40
and it actually launched the time sharing industry because
it was the only one that actually really worked
and gave very good value
for what it cost and it
also was adopted by the ARPA community
so here's angle Bart now a few years after
his proposal because Licklider funded him
giving this incredible demo
called the mother of all demos in 1968
in which pretty much everything that we
have today plus a few things that we don't have today
we're shown so this picture was taken back in
at once and talk to each other with video that was shown in 1968 and it was done on
this project genie machine
and of course the
ARPANET was done right around then
so here's Khan and Cerf who
later got the Turing award for the Internet
but they were heavily involved in the
in doing the ARPANET university
of utah was one of the first few nodes
here had this int machine which
was invented by guess who Wes Clark again
how he has never gotten
the Turing Award I do not know except they don't like Hardware
people but this IMP
was West's idea and today we would call it a router
the idea was you need to have routers
because they
can figure things out they can be
independent of the different kinds of hardware that they're connected to and so
forth this is Lenny Klein rock as he was back
then okay
and then Dave Evans came here in 1965
I came in 66 so
I'm just going to pick up my impressions so
the first time I saw Dave he was wearing
a polo shirt remember
this is in the days where everybody wore you know white shirts
eyes and I couldn't find
a picture of Dave in his polo shirt that's all he wore
except when he had I guess when they took pictures of him he
he was a 44 back
then looked like he was 25
or so and
when I walked into his office he said take this
and read it and what this was was Ivan Sutherlands thesis he
had a big stack of them on his desk and the
new graduate students had to explain
this to him the next day before they got a desk
so and to me that was the biggest
life changer I had because I really
as a programmer for four
or five years beforehand I
really didn't know anything all I knew how to do
is program a computer and sketchpad
as you saw once you see it you realize oh wait a
minute it's not that
computers can do computer like things the computers
can do anything you can actually make
a computer out of a computer that is not at all like the
computer that it's on that's what was interesting about sketchpad
there was nothing about the tx2
that was remotely like the program
that manifested sketchpad and that means
something really incredibly important
Dave was
his it was a bit of a ploy I
discovered later but his his basic idea
was to be invisible this guy was the
greatest motivator the greatest make
things happen around him person I've ever met I
tried to learn from him and
what was funny is that he
unless he was in a
really important stressful situation
where he would grow like 10 feet tall
but I want
to saw that happen a few times rest the
time he'd be sort of well you remember Dave right so he'd
be sort of chuckling and you kind of stammer a
little bit and everything else and meanwhile things are
just sort of happening kind of miraculously another
great line he had when they were complaining about Barton to him
who was a quite a character
to put it mildly he said to the fact the
fact we don't care if they're prima donnas as long as they can sing so
that was the end of that meeting
talking about Ivan he said now
there's a smart citizen to the graduate
wait until they publish get on a plane so I
flew 140,000 air miles
in my two and two years
and three months as a graduate student here because he
travel budgets for all of his graduate students and we
were constantly traveling to both cut not a lot was
happening out in the middle of the country and far as computing
was happening so we would go to both coasts
and meet all the other graduate students meet all
the other principal investigators and this friendliness and cooperation
allowed us out here
hundreds of miles away
from anything to participate fully in the ARPA
community and then fun thing
was one day said we're almost out of money got to go get some more
and so he took a
graduate students along to watch it happen he
would take us out to when we went
to the Pentagon and we could see what it meant to pitch somebody
there for a few million bucks
because he said this is the real world
and what
Chris was talking about earlier Dave said
a PhD is two years of world-class quality
work if you're working
easy project you should finish if you're working on a hard when you don't
have to just write up a progress report and
Ivan was once asked well
what is a PhD thesis and
Ivan said well it's something that three people will sign
barb Barton completely opposite personality
enormous man Dave was
little Barton was enormous kind
of guy who wins push-up contest with Australians
and he
was incredibly articulate so he's kind
of a left-wing version of William F Buckley
and very literate he
was extremely well-read he was a mathematician he
did not like graduate students
he's not really a professor Dave he
was a computer designer it burrows and
Dave convinced him to spend a few years
here and
so here it comes some of his quotes
one of his quotes of a quote that he loved
we should all share in
the excitement of discovery without vain attempts to claim priority
that was this guy in spades so
he could it was very hard to get him to talk about his
own work systems programmers or high priests of a low
cult that's a good one then
in the first class I taught with him it was on advanced
systems design and he came stomping into this class
and he had a sheet of
paper in his hand and says a few things known about advanced
system design they're all written down in these papers I expect
you all to to read them all and understand
them completely and then he said but it is my job to firmly
disabuse you disabuse you of any
fondly held notions you might have brought into this classroom
and so instead of teaching us what
he did is he destroyed us so
anything we believed in with regard to computing he would knock it
down and we would bring up things
that we believed in from his own work and he would knock that down
was total destruction and
because he said look it's not a religion and
you it's
true that some of these things are useful but you
have to you have to start from something
like scratch and you have to admit
these ideas carefully because they will actually rule your
your life so this is best course I ever had in
20-some odd years of school
because he did us the service that most professors
won't do which is to make
us actually bona fide practitioners
and critics in the very field that
we're in something to think about when you're teaching a
class and then this beautiful
one which is used every book everywhere
principle the basic principle of recursive design
is to make the parts have the same powers as the halls and
this works in nature works
in engineering works in
many different ways so so
we had these two guys who are completely different and by the way
I showed up I was graduate student number seven here and
Dave and Bob were
two out of the three professors we had here so
three professors and seven graduate students
and Dave had come here with
some ARPA money because he thought
instead of trying to solve the hidden line
problem you can sort of see what the hidden line problem is by
looking on the scope there and back of John Warnock there's
you draw the edges of things
they show through and the
question is is what if you wanted the surfaces
as though they're opaque what do you have to do and it
was basically an exponential problem in
the early 60s take
minutes and minutes and minutes to just do a
thing with a few hundred surfaces so Dave
was quite sure that continuous-tone graphics using
pixels on a bitmap screen would
actually solve the problem now I have
to tell the story of John here
so John
had a master's in mathematics and he was a very
good programmer and he was working as a staff
programmer in the Merrill
engineering the computer center on
the 1108 he was doing like COBOL
programs for scheduling classes and stuff
like that and not trivial problems
but you know he was a workaday programmer
there in one day one of the graduate
students from the three official 3d project went into John's
office to ask him about handling large arrays and
John said well how large an array and the guy
said well 300,000 misses three hundred
thousand pixels as a is about 640 by
480 300,000
that was a lot in those days and John
said well why do you want to do that and so in
the middle of this explanation of well we're doing this with
graphics John thought of how to do it really how
to really do it and
so he started wrote a Fortran program
to produce these things and
this is one where the power of recursive
to have the parts have the same power of the holes it was a
recursive descent program that turned an exponential
problem into an N log n problem
and he was I think the shortest graduate student we've
ever had here at Utah because
he was promoted to be a graduate student for all of
six weeks I believe and his
thesis was I think probably the thought shortest one still
right 33 pages is like 25 pages of
prose telling
well here's how I did it and here's nine pages of pictures
proving I did it and that was it so
of course that the side story
here which John would never say but it was
because John had this breakthrough that Ivan was willing to come
out here and I've been coming
out here changed the entire course of what
YouTube because it wasn't just one first class guy
and Dave Evans there's one of the towering
geniuses our field is ever had brought himself
in all of his graduate students out here in mass
to to make this department
double triple in size and
gain the stature that it did was because of
John's breakthrough
and I did a few things here
too this is a
project that Dave
was kind of the main mentor
on I mean I Bart and wasn't really a mentor
you could go out and drink beer with him and
so Barton would actually
had the attitude you have to have to
do hard problems that was what was
tremendous about him you just had to be like him and
you could knock it off and so
nothing Dave did it was he got his graduate students
consulting jobs in industry anybody still do that here
yeah and the reason he did it was he
well industry is the real world and graduate
student is an unnatural state they
paid us almost nothing and I asked
him which I said huh why are you paying us so little he says because I don't
want you to stick around he here as a graduate student I want you to
get that degree and get out of here
so this was a so
Dave took me out to meet Edie Cheadle at a
company that was probably no longer exist but it was part of
the link em kovat complex called mem
Corman tech not on the way to the the airport
was an aerospace company and Cheadle was Rollie pollie
Texan who was a genius electrical engineer
and he wanted to do a little desktop
machine and we hit it off and so
the result of that was this thing called the sketch machine it's a
flex machine and you can see it's self-portrait
what it looked like on its own
display the actual prototype looked like that and
it had multiple clipping
windows that had the first object-oriented
user interface
operating system it
had an iconic GUI so it had a lot of the elements of
we're familiar with today and not actually kind of
in 68-69
it looks a little bit like the Apple
computers of ten years later look a little
bit like an apple to there so we
did that and then after
one of these trips to talk
Seymour Papert who is working with children I started thinking
about children's computers and what was cool about the
Flex machine is he could count the transistors in it and
so it's thinking on the plane ride
back Wow I wonder
if you could actually
something that children could carry around because you don't want to children
this would be sitting at a desk so he starts so I sketched this
cartoon of two children
playing a game of space war
which they programmed themselves on
wireless network tablets
and after I got back
here to Utah made this cardboard model and
made it Hollow so
I could fill it up with lead shotgun pellets to see how heavy
you could make it before this is too heavy
by the way we knew that back then so the deal
was about a kilogram is
where you like to be with the
thing and four pounds is the absolute max than
anybody could stand of course you
know these six pounders have been around for a long
time and part of the strength of this was another visit
hat I took to that can't really see that very
well because the the projector is blooming but that
is an inch square flat screen display
from 1968 and
now how did we come to see
that well there was an RPR
Contractors meeting at alt Alta
Lodge in 68
Davis of course took his graduate students along
to this on the ground as long as we didn't
say anything so we've got to watch
all of the leaders of the this powerful
computing community dealing with each other Davis
Dave wanted us to see how the politics worked
and so this is all very interesting all
the stars were there and at the end of
it Bob Taylor
who is the Thunderer at that time
asked us graduate students who are sitting
in a ring around the the back of all of these
famous people if we had any comments and John
Warnock raised his hand and said well
you know we actually do all the work and
we when
we're getting our PhDs and so we're getting out
in the world and you know we do
everything and don't you think there should be something
like this principal investigators meeting
for our graduate students and Taylor
said yes we'll do it this
summer and so that led to many years of
these were the top two graduate students from each place we're chosen
to go to Illinois to
their resume where they did a comp have
a conference center for the University there and span
five days showing
each other what we're doing and forging bonds at left
so this is what that community was like and during that
we went on a field trip to the University of Illinois and here
was this first inch square flat-screen
plasma panel and so you can
you've decided you want to do a computer that can
be carried around and you know how many transistors a
desktop computer has and you know what
Moore's Law is the things
are going to get better over time then you can ask
yourself when is it going to be
that I can put those transistors on the back of a display and
now I will have a tablet
and the answer in 68
was yeah probably around 1980 that was a good
good answer because there's a men
amount of design and invention that had to be
done to to do the
software and the user interface to make these ideas work
okay so now
we're into the 70s and many
of these people that you've met wound
up going to Xerox PARC why because the
US government in its bumbling ways
things responded to some
of the protests about the Vietnam War by curtailing
all government funding on campus including
including the benign ARPA funding that's
when the D was put on our per to make it DARPA and
they've never been the same since and Taylor
who saw what that was
going to means gathered up his favorite graduate
students now who had PhDs and said we kind
of finished this off and so he found Xerox he
was willing to take us on for a few years to
do it and so in a few years four years at Xerox
we had this thing
so this is a 1973 here
like a Macintosh of 1988
or 1989 here's the GUI
desktop publishing
we call real loop these days post
script pages
second laser printer Ethernet
pierre-pierre and client-server and about
Internet committee
and so this all costs in
today's dollars about 10 million dollars a year
so pretty much any place could do
it was done by a grand total of a about 30 people
right so this is cheap gotta
have the right 30 gotta have
the right process we had a legal agreement
with Xerox that they could not interfere with our research
in any way for five years we have to use
that legal agreement because of course they thought
they had ideas also
and here's the interesting thing xerox
made a factor of 300 on
their total investment in Xerox PARC just
from the laser printer alone so contrary
to the urban legend which is not true that Xerox didn't
make anything they paid for Park hundreds of times over
made billions of dollars from the laser
printer they missed the other things but still a return
on investment of you know thirty thousand percent
is pretty good I'm told
the other thing to realize from the standpoint of
the US and the world the return to date from Xerox
PARC is in excess of thirty three trillion dollars
it's an excess of
a trillion dollars a year still just from these
inventions and
so an interesting thing
to contemplate is how come no
government nope
university no country no company will
put out that ten million dollars to fund this way
to think about okay so let's
pass out some hero Awards here so
first Mandarin of course the
Ivan was everything Ivan was
a mentor he was the second funder
he was one of the greatest scientists
we've had these are the mentors who made a difference
Utah in the early days and we give
Dave Evans the main award
because it was his personality that
allowed everything else to function
he could deal he loved crazy people
he knew exactly how to deal
with them and to keep other people from going nuts
because there are crazy people around so he
cannot be helped thank too much and I
tell you if any of his if you wanted to
crossfire any of his graduate students would lie down in it for
him and I'm not kidding
and the grand awards here
are the funding and
the funders these are the four funders of the
ARPA community Licklider Sutherland
Bob Taylor and Larry Roberts
the guys who get domain
awards were both psychologists
Taylor did both funded
the ARPANET and
also it was the guy who set up Xerox PARC
and I believe that
the difference between now and then is as
simple as the difference between how
funding is done the funding then attracted
a particular type of people those people
are still around today but they're there's a complete
mismatch between them and the funding so those people are going into something
else other than computing I
myself would not go into computing today if
I were 25 it's
almost impossible to go to a university and even find
out that it's neat because it's so
intertwined it's been so invaded by IBM ISM but
now in the form of Microsoft and web
and everything else so there's been a complete confusion
between what it means to be a vocational
person in computing and what it means to actually get an education
in and the universities have been generally
the the biggest
sinners in that because universities
with when the baby boom came along pretty much turned themselves
into businesses which they pretty much are today
ok a little joke well
I'm not going to do the joke I'm just going to put those up as an enigmatic
thing here
because I think this is a good time to pause for any questions you
might have because I've done I've done a bit just about an
hour and I think that's all I should should do here
I'll leave the joke until later so
any questions on
this talk which is
mostly about why I think
things worked and why I don't think
they work as well today thank
a few questions before the reception yes sir
did the ARPANET help to cross fertilize
these ideas that you talk with others in other places
in ways that were interesting that made things happen
yes it did the
ARPANET cross-fertilize ideas yes
and interestingly
just as much before it was
built as afterwards because
it was these several of these large projects
that require cooperation
between these 15
places got
a lot of the cross fertilization happening in
doing the planning and it took
some years before the ARPANET was
actually as you you know email
existed before we
networks that existed on the time sharing systems
of the 60s and
everybody was very very happy to be
able to have a working email
system that linked colleagues
together and so the
but I think that the you
know for instance FTP file sharing
and other kinds of things were done after the
advent of the the ARPANET which is really an experiment
in adaptive
packet switching many ways the ARPANET
was a little bit more interesting than the way the internet
does it because it actually it
did the ARPANET actually thought to a certain extent
so there's a lot of fun working
in many of us like I was on the ARPANET committee
I was one of many graduate students
who served on that yeah
well I think the
the you know I was
willing to settle for Less hardware-wise than
we have today but to me to me the Dynabook was
always a service idea this is something
that most people have still have a hard time understanding about
computing because they you
know we still place a lot of
value in atoms it's
hard to actually monetize things unless
you get something in the mail right
or you find a very low price for it because
don't know how to value those things that
computers are actually undervalued as far as I'm concerned
the average American car cost twenty
eight thousand five hundred and I pay that in a flash
for twenty eight thousand five hundred dollars worth
of computing in a box like this rather than 2k
right but most people don't use the computer
for anything useful it's just no
it's true because people almost
everybody uses the computer almost all the time as a convenience
for dealing with old media
hardly anybody of the billions
who use computer learns about new
things by writing simulations for example so nothing
even remotely resembling a computer computer
literacy exists so this is bait
this is still the Gutenberg Bible imitating
the fonts and
ligatures that the monks did
and a good way to think about computers is to
services so there are actually three physical forms I
came up with for the Dynabook they
were rather different from each other based on
things that were going on but the you
know most of the thinking I did about it was what
services and part of the service
ideas what kind of user interface do you actually
have to have to help
people rather than hinder them and it took us
while at Xerox PARC to come up with one that
was even halfway decent that's the one that's used today
and and it was really a much better
interface for children than it than it has been for adults
I think yeah so
if you look at the services provided by something
like an iPad they're terrible right
because it violates the first principle of
personal computing which is symmetric creation with
consumption so that was one of the
few notice and sketchpad you
with a working bridge or you could make one and
microsoft word which was originally
done at Parc you could present somebody
with a document and you could read it or you can make a document
but notice that symmetry exists
iPad the iPad most of the stuff that Apple has been doing
our consumer consumption devices
and I think they're terrible for people
because they can completely gotten
right price so that people will
buy an app for whatever
it is rather than improving things enough
so that they can make tools as they need
them right so that's anti what this
whole romance of personal
computing was about from this group yeah
so and you know whenever you make something
whenever you make a tool you're you're simultaneously
making an amplifier and a prosthetic
car amplifies us in one way but it
withers our body in another we have
to choose to exercise once we take on a
car and
Socrates you
know complained about writing that it takes away the
need to remember things and
of course this was Plato
being ironic because
Plato was the one of the first great writers of all
time and he would use Socrates as his mouthpiece
so I'm sure he was chuckling like mad while he
was while he was writing this and he didn't
say it but anybody with a brain would
realize oh but wait a minute I don't have to give
up my memory just because I've got writing now
I've actually got the best of both worlds
providing I decide to remember because I
certainly don't want to leave anything interesting out in a manuscript
right it's too inefficient it's
too inefficient to try and find things
you need them on this thing so you might as well when you
learn how to read you might as well learn how to remember and now you're
really powerful because reading is many times more
efficient than oral
transmission and if
you can remember you're now putting enormous
amounts of stuff that can work with each other inside
your head so that was
so these are the kinds of thinking that we were
doing back back then
yes sir
today's money
you know I've often wondered how
how much of a simple generalization you can
make and get away with on
this but one thing that is certainly true
one contrast that is really
noticeable is that today's
funders like the
funders in the past are responsible they have this money
but today's funders
confuse being responsible
with the need to control and the
funders in the past absolutely did not
that's so
look lighter didn't control the darn thing his
idea was in fact so a
good Licklider story is they
asked him about
failure is
any worry because if you looked
at what these different projects were trying to do they were cosmic
every one of the things had not
there was nothing like them that had actually
been done before so pretty much it was artificial intelligence
it was making a network that could
scale by 10
magnitude without breaking you know nobody done anything
like that Licklider said well we're not playing golf
we're losing a stroke is a
tragedy so this is more
like baseball and Tyco Ty
Cobb had the best lifetime batting average and
it was 367 so two-thirds
of the time you know something
not good
happened when Ty Cobb went to the plate and
if you look at most of the assessment systems that are
set up now by Congress
and in businesses people who fail two-thirds of the time
or gotten rid of but in sports they
understand what the deal is and
Licklider said look if we if we're 30
or 40 percent successful on what
we're funding just that will change the entire world
and they said well what about the 60 or 70
percent I said well that's just the cost of doing business in research
and this is what people don't understand because
this is a phrase that businesses do have
for various ephemeral things like advertising
but they have never been able to apply it
to what long-range research does
in long range research has been incredibly
fruitful but another
reason you could say is down deep these
people don't want to fund anything they don't understand
right and that
is a disaster because it's not their business to understand
that's what they're paying the scientists
for their business is to find
the best people they can and give them the money and take the percentages
on the results if they do that well you'll
always get multi trillion-dollar returns
on the thing so a lot of different reasons that are
somewhat similar right you can work them in
different directions part of it
is Congress was not a factor back
then partly because of Proxmire
in the seventies and other things
happened since Congress has inserted itself very
very deeply in
these Affairs including this turning
ARPA into DARPA which came with a
congressional oversight that makes that requires DARPA
to have goal-oriented proposals
which means
you have to you have to tell the funders what you're
going to do ahead of time and that's not long-range research
long-range researches the
funders are paying partly for problem finding
so I've been working on NSF for the last few
years to get this say look you guys can't keep funding engineering
proposals and getting it you're not doing anything
you know you're you're keeping professors and
graduate students alive but the field is dying
because nothing is happening
and it's not that these breakthroughs we're
and now there aren't anymore breakthroughs and it's
people back then we're any more special than they are today
numerically there are more people
of the caliber of ARPA and Xerox
PARC today than they were back then I mean the field is enormous
so there's more to draw from and
Jesus the computing just
just having
lots of memories
to work with is something
that we would have died for back then
so so
there aren't so this is like a poco Barrett you remember
poco he said we've met the enemy and they are us
this is one of these things where this
is the simplest proposition and
another thing that is astounding to
me is that when
they damped demand rationalization of these things I said
look the best rationalization for investing
is portfolio investing it's proven
mathematically then in portfolio investing
you have to invest some small percentage in
unvetted projects
because you'll get mat the
maximum return that way and if
you go to universities you go to businesses they learn
this in MBA school and they know it yep
yep yeah but they won't do it they want that small
percent to go into their bottom line because
they don't understand so
it's it has to do with the
the different the difference between I
think the regular world and things
that require talent like team
sports right
nobody's going to demand democracy in professional
all right you get the best they
millions for them a professional baseball and you're
not looking for the same kind of person your
guts the whole point of a team a team is
a multi-faceted diamond that
is put together partly by the coach and partly
by the way the players work off and
whole point is to get synergy so it's not about having
a a party line or a
religion not about any of these things that make
people feel comfortable you know and the
as Licklider said you know what
do we don't worry but like the part
tailor was asked wasn't this going to be expensive and
Taylor said no because I'm not going to hire any good
person and Zurich said what and
he said no because you can't concatenate good people
to do what a great person can do says I'm
only got hire great people and there are that many
of them so the budget is limited
but he says once I have these
don't have to have a management structure with
these people because these people already know what they want to do
so Taylor said my job at Xerox PARC is
to set up the social conditions so
when these lone wolves need to cooperate with each other
they will that was what his job was he
never gave us a directive or suggestion
on any goal we should work on but he worked on the
cology of the system that was his role in the thing
Butler was the brilliant Oppenheimer we
had my group was the lunatic fringe as it
was called and and
we didn't have a party line on the on the thing and we
did cooperate when that when that was a good
idea and it was powerful like two dozen people you
we just did a lot of things in four years and
people are amazed at and I said no with it that was the easiest
hing we ever did in our life it's the one time we ever got to
interference from people who don't know what they're doing
so that's the that's the
hell of it this stuff is really easy
to pull off if
you don't have all these side conditions that
wind up killing things before they ever get started